From Bits to Rounds: Parallel Decoding with Exploration for Diffusion Language Models

This research introduces Explore-Then-Exploit (ETE), a training-free decoding strategy that leverages information-theoretic principles to significantly reduc...

Level: advanced

By Hengyu Fu, Baihe Huang, Virginia Adams, Charles Wang, Venkat Srinivasan, Jiantao Jiao

Category: research